Dear "David Andrey",
Please do not top post / full qoute. Read http://www.netmeister.org/news/learn2quote.html
In message 1OT8ED-000558-1r@wolf.netmodule.com you wrote:
he NAND is formatted as follow:
- U-Boot Partition
- Images Partition (YAFFS2) (Kernel, RootFS as tgz, FPGA / DSP
Firmwares).
- Linux RootFS (YAFFS2)
To update "Linux", the way we imagine is this one:
- write a new kernel and rootFS in the "Images" partition (get the
files through TFTP) 2. u-boot erase the old rootFS and "install" the newer 3. Boot Linux with the newer kernel
If you have a network connection an can download the images through TFTP, then why do you need the "Images Partition" at all? You coulinstall the downloaded images directly.
But why from U-Boot ? :-) -> To provide a robust update mechanism. -> equals what happens with Linux (through end-user), u-boot will "always" start.
OK.
So, I'm looking to clarify the step 2. What is missing ? Ideally, something like this: yrdm /images/my_newer_rootfs.tgz my_ram_address yunzip my_ram_address /rootfs/
For me the reason to have the images partition is missing. If I were in your place, I would probably use the spoace rather to have space for a second root file system, so I can always install into an alternative partition while keeping the old (working copy) in place. Even if an update fails permanently for some reason (like corrupted images on the server) you can then still fall back to the old version.
And as long as you download the images anyway there is little reason to store these locally in a non-installed form. At least I cannt see one.
The patch's I means are: http://lists.denx.de/pipermail/u-boot/2008-September/040910.html http://lists.denx.de/pipermail/u-boot/2006-September/016867.html
That's old and obsolete stuff. Should not be needed at all.
So, what do you think ? Sense or no sense ? Is it a way to achieve this from u-boot ?
It can be done as you described it, but it makes little sense to me.
Best regards,
Wolfgang Denk